REVIEW 4 cited by
The limits of distribution-free conditional predictive inference
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
We consider the problem of distribution-free predictive inference, with the goal of producing predictive coverage guarantees that hold conditionally rather than marginally. Existing methods such as conformal prediction offer marginal coverage guarantees, where predictive coverage holds on average over all possible test points, but this is not sufficient for many practical applications where we would like to know that our predictions are valid for a given individual, not merely on average over a population. On the other hand, exact conditional inference guarantees are known to be impossible without imposing assumptions on the underlying distribution. In this work we aim to explore the space in between these two, and examine what types of relaxations of the conditional coverage property would alleviate some of the practical concerns with marginal coverage guarantees while still being possible to achieve in a distribution-free setting.
Forward citations
Cited by 4 Pith papers
-
Conformal prediction with localization
Localized conformal prediction reweights training scores by proximity to the test point and adjusts the quantile level to preserve distribution-free finite-sample coverage.
-
Decision Theoretic Foundations for Conformal Prediction: Optimal Uncertainty Quantification for Risk-Averse Agents
Prediction sets paired with max-min decisions are a sufficient statistic for value-at-risk optimizing agents, and the RAC algorithm builds such sets with distribution-free coverage.
-
A Unified Comparative Study with Generalized Conformity Scores for Multi-Output Conformal Regression
New CDF-based and latent-space conformity scores give multi-output conformal predictors asymptotic conditional coverage while retaining finite-sample marginal coverage.
-
With Malice Towards None: Assessing Uncertainty via Equalized Coverage
A conformal prediction wrapper that guarantees each protected group's prediction interval covers the true value at the target rate, using only an exchangeability assumption.
Discussion (0). Continue with ORCID to comment.